-
fastllm Public
fastllm是后端无依赖的高性能大模型推理库。同时支持张量并行推理稠密模型和混合模式推理MOE模型，任意10G以上显卡即可推理满血DeepSeek。双路9004/9005服务器+单显卡部署DeepSeek满血满精度原版模型，单并发20tps；INT4量化模型单并发30tps，多并发可达60+。
-
-
reedsolomon-c-neon Public
Forked from jannson/reedsolomon-cC compatible version for https://github.com/klauspost/reedsolomon. with neon
C++ MIT License UpdatedOct 16, 2023 -
-
mongoose Public
Forked from cesanta/mongooseMongoose Embedded Web Server Library - Mongoose is more than an embedded webserver. It is a multi-protocol embedded networking library with functions including TCP, HTTP client and server, WebSocke…
C Other UpdatedMay 23, 2020 -
-
-
-
libnfporb Public
Forked from kallaballa/libnfporbImplementation of a robust no-fit polygon generation in a C++ library using an orbiting approach
C++ GNU General Public License v3.0 UpdatedMay 22, 2018 -
caffe Public
Forked from NVIDIA/caffeCaffe: a fast open framework for deep learning.
C++ Other UpdatedJul 6, 2017 -
-
-
{{ message }}